JP6876368B2 - User authentication method and device using biological signals - Google Patents

Description

The present invention relates to a user authentication method and device using a biological signal.

Technologies for utilizing various signals and data extracted from living organisms in various systems are being developed. In particular, biometric authentication technology that builds security systems using biological signals and data is in the limelight. The biometric authentication technology is a technology that extracts signals and data related to a living body from a user, compares the signals and data with the data already stored, and confirms the identity of the user to authenticate the user. Typically, as one of the fields of biometric authentication technology, a technology for authenticating a user using an individual's electrocardiogram signal has been developed.

Since biometric authentication technology uses an individual's unique biological signal, there is no risk of theft or loss, and it is difficult to forge or alter it, so it is in the limelight in the security field.

The present invention has been made in view of the above prior art, and an object of the present invention is to provide a user authentication method and device for accurately authenticating a user using a biological signal.

The authentication device according to one aspect of the present invention, which has been made to achieve the above object, has a data set generator that extracts a plurality of waveforms from a user's biological signal to generate an authentication data set, and each of the extracted waveforms. A similarity calculation unit that matches a plurality of registered waveforms contained in a registered data set stored in a memory and calculates the similarity between each of the extracted waveforms and the registered waveform. The representative authentication waveform (representative assistance waveform) indicating the representative waveform of the extracted waveform and the representative registered waveform (representative registration waveform) indicating the representative waveform of the registered waveform are extracted and the representative authentication is performed. A preliminary similarity calculation unit (auxiliary similarity waveform) that calculates the similarity between a waveform and the representative registered waveform, and the similarity between each of the extracted waveforms and the registered waveform, and the above. It is provided with an authentication unit that authenticates a user using the similarity between a typical authentication waveform and the representative registered waveform.

The biological signal may include an electrocardiogram signal.
The memory may store a registration data set containing a plurality of registered waveforms extracted from a user's biological signal during registration before the data set generation unit generates the authentication data set.
The similarity calculation unit extracts each feature of the extracted waveform and each feature of the registered waveform, and has each feature of the extracted waveform and each feature of the registered waveform. The similarity between can be calculated.
The similarity calculation unit converts each of the extracted waveform and the registered waveform into a feature vector, and each of the feature vectors converted from the extracted waveform and the registered waveform are converted. The degree of similarity between each of the feature vectors can be calculated.
The similarity calculation unit may change the dimensions of the extracted waveform and the registered waveform to convert each of the extracted waveform and the registered waveform into the feature vector.
The similarity calculation unit extracts each feature parameter of the extracted waveform and each feature parameter of the registered waveform, and each feature parameter of the extracted waveform and each of the registered waveforms. The degree of similarity with the feature parameters of can be calculated.
The characteristic parameters are the PR interval (PR interval), PR segment (PR segment), QRS complex (QRS amplitude), ST segment (ST segment), and ST interval (ST segment) of the extracted waveform or the registered waveform, respectively. It may include at least one of ST segment, QT interval, RR interval, amplitude, or noise power.
The preliminary similarity calculation unit extracts the characteristics of the representative authentication waveform and the representative registered waveform, and the similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. Can be calculated.
The preliminary similarity calculation unit converts each of the representative authentication waveform and the representative registered waveform into a feature vector, and the feature vector converted from the representative authentication waveform and the representative registered waveform are obtained. The degree of similarity with the transformed feature vector can be calculated.
The preliminary similarity calculation unit extracts the characteristic parameter of the representative authentication waveform and the characteristic parameter of the representative registered waveform, and the characteristic parameter of the representative authentication waveform and the characteristic parameter of the representative registered waveform. The degree of similarity between and can be calculated.
The authentication unit has a degree of similarity between each of the extracted waveforms and each of the registered waveforms, and a degree of similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. Can be used to determine if a user matches a user in the registered dataset.
The authentication unit has a degree of similarity between each of the extracted waveforms and each of the registered waveforms, and a degree of similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. Can be normalized.
The authentication unit has a degree of similarity between each of the extracted waveforms and each of the registered waveforms, and a degree of similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. Can be set differently to determine if the user matches the user in the registered dataset.
The authentication unit has a similarity between each of the extracted waveforms and each of the registered waveforms in the classifier, and between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. It is possible to determine whether or not the extracted waveform and the registered waveform correspond to each other by applying the similarity of.
The classifier may include at least one of SVM (Support Vector Machine) or NN (Nearest Neighbor).
When the authentication unit determines that the user does not match the user of the registered data set, the authentication device of the data set generation unit, the similarity calculation unit, the preliminary similarity calculation unit, and the authentication unit. The operation can be repeated.
When the number of repetitions of the operations of the data set generation unit, the similarity calculation unit, the preliminary similarity calculation unit, and the authentication unit exceeds a predetermined critical number of repetitions, the authentication unit causes the user to use the authentication unit. It can authenticate if it does not match the user in the registered dataset.

An authentication device according to another aspect of the present invention, which has been made to achieve the above object, has a sensor that senses a user's biometric signal and extracts a plurality of waveforms from the biometric signal to generate an authentication data set, which is extracted. Each of the waveforms is matched with a plurality of registered waveforms stored in the memory, the similarity between each of the extracted waveforms and the registered waveform is calculated, and the representative of the extracted waveforms. A typical authentication waveform indicating the waveform and a representative registered waveform indicating the representative waveform of the registered waveform are extracted, and the similarity between the representative authentication waveform and the representative registered waveform is calculated. , A processor that authenticates the user using the similarity between each of the extracted waveforms and the registered waveform and the similarity between the representative authentication waveform and the representative registered waveform. To be equipped.

The biological signal may include an electrocardiogram signal.
The processor extracts each feature of the extracted waveform and each feature of the registered waveform, and between each feature of the extracted waveform and each feature of the registered waveform. The similarity can be calculated.
The processor can extract the characteristics of the representative authentication waveform and the representative registered waveform and calculate the similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. ..

The authentication method of the user authentication device according to one aspect of the present invention, which is performed to achieve the above object, is a step of extracting a plurality of waveforms from a user's biological signal to generate an authentication data set, and each of the extracted waveforms. Is matched with a plurality of registered waveforms contained in the registered data set stored in the memory to calculate the similarity between each of the extracted waveforms and the registered waveform, and the extracted waveform. A typical authentication waveform showing a representative waveform of the waveform and a representative registered waveform showing the representative waveform of the registered waveform are extracted, and the similarity between the representative certified waveform and the representative registered waveform is obtained. The user using the step of calculating the degree and the similarity between each of the extracted waveforms and the registered waveform and the similarity between the representative authentication waveform and the representative registered waveform. Has a step to authenticate and.

According to the user authentication method and apparatus of the present invention, the influence of the highly specific waveform of the authentication data set including a plurality of authentication waveforms or the waveform of the registered data set including a plurality of registered waveforms on the user authentication. By reducing the number of false recognitions, the false recognition rate and the false rejection rate can be reduced, so that the user can be accurately authenticated using the biometric signal. In addition, not only the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, but also the similarity between the representative authentication waveform and the representative registered waveform is displayed in the classifier. By applying it, the false recognition rate and false rejection rate can be reduced.

(A) is a block diagram showing an authentication device according to one embodiment, and (b) is a diagram for explaining an example of an electrocardiogram signal. It is a block diagram which shows the authentication apparatus by another embodiment. It is a block diagram which shows the authentication apparatus by still another Embodiment. It is a block diagram which shows the authentication apparatus by still another Embodiment. (A) and (b) are diagrams for explaining user authentication of the authentication device according to one embodiment. It is a figure for demonstrating the calculation of the degree of similarity between the extracted waveform and the registered waveform by one Embodiment. It is a figure for demonstrating the matching of the extracted waveform and the registered waveform by one Embodiment. It is a figure for demonstrating an example of the authentication apparatus by one Embodiment. It is a figure for demonstrating an example of the authentication apparatus by another embodiment. It is a figure for demonstrating an example of the authentication apparatus by still another Embodiment. It is an operation flowchart which shows the authentication method by one Embodiment.

Hereinafter, specific examples of embodiments for carrying out the present invention will be described in detail with reference to the drawings. The same reference numerals presented in each drawing indicate the same members.

Various modifications can be made to the embodiments described below. The embodiments described below are not intended to limit this embodiment and include all modifications, equivalents or alternatives to these.

The terms used in this embodiment are used solely to describe a particular embodiment and are not intended to limit this embodiment. A singular expression includes multiple expressions unless they mean something that is clearly different in context. In the present specification, terms such as "including" or "having" mean that the features, numbers, steps, actions, components, parts, or a combination thereof described in the specification exist. It does not preclude the existence or addition of one or more other features or numbers, steps, movements, components, parts, or combinations thereof.

Unless defined as different, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by those with ordinary knowledge in the technical field to which this embodiment belongs. Have. Commonly used predefined terms are to be construed as having a meaning consistent with the meaning they have in the context of the relevant technology and are ideally or overly formal unless expressly defined herein. It is not interpreted as a meaning.

Further, in the description referring to the drawings, the same reference reference numerals are given to the same components regardless of the drawing reference numerals, and duplicate description thereof will be omitted. If it is determined in the description of the present embodiment that the specific description of the related known art unnecessarily obscures the gist of the embodiment, the detailed description thereof will be omitted.

FIG. 1A is a block diagram showing an authentication device according to an embodiment, and FIG. 1B is a diagram for explaining an example of an electrocardiogram signal.

Referring to FIG. 1A, the authentication device 100 includes a data set generation unit 110, a similarity calculation unit 120, and an authentication unit 130. The data set generation unit 110, the similarity calculation unit 120, and the authentication unit 130 include one or more processors or memories.

The data set generation unit 110 extracts a plurality of waveforms for user authentication from the user's biometric signal, generates an authentication data set, and stores the authentication data set in the memory. Here, the biological signal indicates a signal sensed from the human body, for example, the biological signal is an electrocardiogram (ECG) signal, an electroencephalogram (EEG) signal, an electromyogram (EEG). Includes Electro Micro Myo Gram (EMG) signal, electrocardiogram (Electro Oculo Gram: EOG). In addition, the biological signal may include other biological signals other than those listed above. The data set generation unit 110 receives a biological signal from a sensor (for example, an electrocardiogram sensor, an electrocardiogram sensor, an electromyogram sensor, an electrocardiogram sensor). Here, the sensor is a hardware device. The sensor is included in the authentication device 100 or is an external device other than the authentication device 100. Further, the authentication device 100 can receive a biological signal not only from a sensor for a specialized medical device but also from a sensor included in a wearable device or a mobile device.

Hereinafter, the present specification will be described based on an electrocardiogram signal for convenience of explanation. Of course, the biological signal used in the present specification is not limited to the electrocardiogram signal, and other biological signals are also used.

In one embodiment, the electrocardiogram sensor comprises a plurality of electrodes, an amplifier, and a digital filter. The plurality of electrodes come into contact with the user's skin (for example, a finger) to sense the user's electrocardiogram signal. The amplifier amplifies the electrocardiogram signal sensed by the plurality of electrodes. In one embodiment, the amplifier is referred to as an analog front end (AFE). The digital filter converts the amplified electrocardiogram signal into a digital signal. As a result, the signal-to-noise ratio (Signal-to-Noise Ratio: SNR) of the electrocardiogram signal is improved.

In another embodiment, the electrocardiogram sensor senses the user's electrocardiogram signal using only a plurality of electrodes.

The data set generation unit 110 extracts a plurality of authentication waveforms for authentication from the electrocardiogram signal. The data set generation unit 110 divides the electrocardiogram signal based on the heart rate and extracts a set of waveforms including P wave, QRS wave, T wave, and U wave. An example of P wave, QRS wave, T wave, and U wave is shown in FIG. 1 (b). Further, the data set generation unit 110 evaluates the quality (for example, SNR) of the extracted waveform, and does not extract a waveform having a quality lower than a preset quality value as an authentication waveform.

The data set generation unit 110 in the present embodiment performs a preprocessing process for removing the main noise of the plurality of extracted authentication waveforms. For example, in the preprocessing process, the data set generation unit 110 extracts only a plurality of extracted authentication waveforms of 0.5 Hz to 40 Hz from the plurality of extracted authentication waveforms. In addition, the data set generation unit 110 performs DC baseline wandering (DC baseline wandering), power noise (for example, an electrocardiogram waveform in the frequency band of 50 Hz to 60 Hz), and motion of a plurality of extracted authentication waveforms. Eliminates major noise such as motion artifacts.

Further, the data set generation unit 110 arranges a plurality of extracted authentication waveforms with reference to the R peak.

The data set generation unit 110 includes a plurality of extracted authentication waveforms in the authentication data set. Here, the authentication data set means a set configured as an authentication waveform of the user in order to authenticate the user. As an example, the authentication dataset is represented as a probe set. The data set generation unit 110 accumulates the electrocardiogram waveform of the user in the authentication data set, and deletes the authentication data set after the user's authentication is completed.

Further, the authentication device 100 includes a memory storage unit (not shown). The memory storage unit (not shown) stores in advance a registration data set including a plurality of registered waveforms extracted from the biomedical signals of the user registered in advance during registration. Here, the registered data set means a data set configured as a registered waveform of the user in order to register the user. In one embodiment, the authentication device 100 acquires the biological signal of the user who is going to register, extracts a plurality of waveforms for user registration from the acquired biological signal, and generates a registration data set. Alternatively, the authentication device 100 extracts a plurality of waveforms from the biological signals of the user stored in the authentication device 100 before performing authentication to generate a registered data set. In another embodiment, the authentication device 100 receives a registration data set from an external device.

The similarity calculation unit 120 performs one-to-one matching with each of the plurality of extracted authentication waveforms included in the authentication data set with the plurality of registered waveforms included in the registered data set, and performs one-to-one matching with the plurality of extracted authentication waveforms. Calculate the similarity between each of the waveforms and a plurality of registered waveforms. That is, the similarity calculation unit 120 matches the authentication data set and the registered data set, and calculates the similarity between each of the plurality of extracted authentication waveforms and the plurality of registered waveforms. As an example, when a plurality of extracted authentication waveforms included in the authentication data set are M and a plurality of registered waveforms included in the registration data set are N, the similarity calculation unit 120 is M ×. Matching is performed N times, and the degree of similarity between each of the M extracted authentication waveforms and each of the N registered waveforms is calculated. Further, the similarity calculation unit 120 generates M × N histograms showing the similarity between each of the M extracted authentication waveforms and each of the N registered waveforms.

The similarity calculation unit 120 does not extract a representative authentication waveform from a plurality of extracted authentication waveforms included in the authentication data set, and also does not extract a representative authentication waveform from a plurality of registered waveforms included in the registration data set. Is not extracted, and the entire plurality of extracted authentication waveforms included in the authentication data set are matched one-to-one with the entire plurality of registered waveforms included in the registration data set. As a result, the loss of information generated by extracting the representative authentication waveform or the representative registration waveform is prevented, and the waveform having a large specificity among the authentication data set or the waveform included in the registration data set is used by the user. The impact on certification can be reduced. For example, if one of the four extracted waveforms contained in the authentication dataset contains a lot of noise, then a typical authentication waveform extracted from the four extracted authentication waveforms contains a lot of noise1 It may be distorted by one waveform. The similarity calculation unit 120 does not extract a representative authentication waveform from the four extracted authentication waveforms included in the authentication data set, and registers each of the four authentication waveforms in a plurality of registered data sets. By matching with each of the waveforms, it is possible to reduce the authentication error due to one waveform containing a lot of noise.

In order to calculate the similarity, the similarity calculation unit 120 extracts the characteristics of each of the plurality of extracted authentication waveforms and the characteristics of each of the plurality of registered waveforms. Here, the feature shows the characteristic of the waveform and is also expressed as a feature point. In this embodiment, features include feature vectors or feature parameters. Also, not limited to this, the feature may include all configurations exhibiting the characteristics of the waveform.

The similarity calculation unit 120 in the present embodiment converts each of the plurality of extracted authentication waveforms and the plurality of registered waveforms into feature vectors. The similarity calculation unit 120 changes the dimensions of each of the plurality of extracted authentication waveforms and the plurality of registered waveforms, and converts each of the plurality of extracted authentication waveforms and the plurality of registered waveforms into a feature vector. To do. For example, the similarity calculation unit 120 includes an AC / DCT (Auto-correction / Discrete Cosine Transform), a STFT (Short Time Forier Transform), an MFCC (Mel-frequency Cepstra A plurality of extracted authentication waveforms and a plurality of registered waveforms are converted into a frequency domain by using a Predictive Spectram Cofficient) or an LSP (Line Spectram Pair). The similarity calculation unit 120 is converted into a frequency domain converted into a frequency domain based on a PCA (Principle Component Analysis), an LDA (Linear Discriminative Analysis), or an ERE (Eigenfeature Migration and Execution). Select each characteristic component of the waveform to reduce the dimension of each of the plurality of extracted authentication waveforms and the plurality of registered waveforms, and reduce the dimensions of the plurality of extracted authentication waveforms and the plurality of registered waveforms. A feature vector is generated from each of the waveforms.

The similarity calculation unit 120 in the present embodiment extracts each feature parameter of the plurality of extracted authentication waveforms and each feature parameter of the plurality of registered waveforms. Here, the feature parameter includes at least one of the PR interval, PR segment, QRS complex, ST segment, ST interval, QT interval, RR interval, amplitude, or noise power of the electrocardiogram waveform.

The similarity calculation unit 120 calculates the similarity between each feature of the plurality of extracted authentication waveforms and each feature of the plurality of registered waveforms. As an example, the similarity calculation unit 120 may be used for correlation, cosine similarity, Euclidean distance, L1 waveform, P-nom, or root mean square error (Root Mean Square). Using Root-mean-square (RMSE), the similarity between each of the plurality of feature vectors converted from a plurality of extracted authentication waveforms and each of the plurality of feature vectors converted from a plurality of registered waveforms, or The degree of similarity between each feature parameter of the plurality of extracted authentication waveforms and each feature parameter of the plurality of registered waveforms is extracted. Here, the larger the value of the relevance or cosine similarity is, the higher the similarity is, and the smaller the value of the Euclidean distance or the root mean square error is, the higher the similarity is. In addition, the similarity calculation unit 120 uses all the methods capable of comparing the features between the features of the plurality of extracted authentication waveforms and the features of the plurality of registered waveforms. Similarity can be extracted.

The similarity calculation unit 120 in the present embodiment determines the similarity between each feature of the plurality of extracted authentication waveforms and each feature of the plurality of registered waveforms as a histogram, a feature vector, or a value. It is shown in the form of.

The authentication unit 130 authenticates the user using the similarity. The authentication unit 130 determines whether or not the user matches the user of the registered data set by using the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms.

In one embodiment, in order to perform authentication, the authentication unit 130 normalizes the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms. Here, the authentication unit 130 uses a normalization function such as min-max, Tanh-estimator, or z-score to be used between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms. Normalize the similarity.

The authentication unit 130 applies the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms to a classifier (not shown), and a plurality of extracted authentication waveforms and a plurality of the same. Judge whether or not it corresponds to the registered waveform of. When it is determined that the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other, the authentication unit 130 authenticates that the user of the authentication data set matches the user of the registered data set, and a plurality of extracted authentication waveforms are extracted. When it is determined that the authentication waveform and the plurality of registered waveforms do not correspond to each other, the authentication unit 130 authenticates that the user of the authentication data set does not match the user of the registered data set.

As an example, the classifier is an SVM (Support Vector Machine) or NN (Nearest Neighbor), and the authentication unit is an SVM or NN with each of a plurality of extracted authentication waveforms and each of a plurality of registered waveforms. It is determined whether or not the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other by applying the similarity between them.

When the authentication unit 130 determines that the plurality of extracted authentication waveforms and the plurality of registered waveforms do not correspond to each other, the authentication device 100 includes the data set generation unit 110, the similarity calculation unit 120, and the authentication unit 130. Repeat the operation. For example, when it is determined that the plurality of extracted authentication waveforms and the plurality of registered waveforms do not correspond to each other, the data set generation unit 110 reacquires the electrocardiogram signal of the user who is trying to authenticate, and the reacquired electrocardiogram. Extract multiple authentication waveforms from the signal to generate a new authentication data set. The similarity calculation unit 120 matches each of the plurality of newly extracted authentication waveforms included in the new authentication data set with the plurality of registered waveforms included in the registered data set on a one-to-one basis, and a plurality of newly extracted waveforms are newly extracted. The similarity between each of the authentication waveforms and the plurality of registered waveforms is calculated, and the authentication unit 130 authenticates the user using the newly calculated similarity. Here, the authentication unit 130 registers the user as a user of the data set when the number of repetitions of the operations of the data set generation unit 110, the similarity calculation unit 120, and the authentication unit 130 exceeds a predetermined critical number of iterations. Authenticate if it does not match.

Due to the operation of the data set generation unit 110, the similarity calculation unit 120, and the authentication unit 130, the authentication device 100 eliminates the loss of information generated by extracting a typical authentication waveform or a typical registered waveform. By preventing and reducing the influence of highly specific waveforms among the waveforms contained in the authentication data set or the registered data set on the user's authentication, false recognition rate (FAR) and false rejection rate (FAR) False Rejection Rate (FRR) can be reduced.

FIG. 2 is a block diagram showing an authentication device according to another embodiment.

Referring to FIG. 2, the authentication device 200 includes a data set generation unit 210, a similarity calculation unit 220, a preliminary similarity calculation unit 230, and an authentication unit 240. The data set generation unit 210, the similarity calculation unit 220, the preliminary similarity calculation unit 230, and the authentication unit 240 include one or more processors or memories.

The data set generation unit 210 generates an authentication data set by extracting a plurality of waveforms for user authentication from the user's biological signal.

The similarity calculation unit 220 matches each of the plurality of extracted authentication waveforms included in the authentication data set with the plurality of registered waveforms included in the registered data set on a one-to-one basis, and performs a plurality of extracted authentications. Calculate the similarity between each of the waveforms and a plurality of registered waveforms.

Since the contents of the data set generation unit 110 and the similarity calculation unit 120 shown in FIG. 1A are applied as they are to the data set generation unit 210 and the similarity calculation unit 220, more detailed description will be omitted.

The preliminary similarity calculation unit 230 extracts the characteristics of the typical authentication waveform of the authentication data set and the characteristics of the representative registered waveform of the registered data set, and sets the characteristics of the typical authentication waveform and the characteristics of the typical registered waveform. Extract the similarity that indicates the similarity between them. Here, the representative authentication waveform shows the representative waveform of the plurality of extracted authentication waveforms included in the authentication data set, and the representative registered waveform is the representative waveform of the plurality of registered waveforms included in the registered data set. Shown. As an example, the preliminary similarity calculation unit 230 calculates the average value of a plurality of extracted authentication waveforms to generate a representative authentication waveform, and calculates the average value of a plurality of registered waveforms to represent the representative. Generate a registered waveform. Alternatively, not limited to this, the preliminary similarity calculation unit 230 can extract a representative authentication waveform and a representative registered waveform by using another statistical method.

The preliminary similarity calculation unit 230 extracts the features of a typical authentication waveform and a typical registered waveform.

The preliminary similarity calculation unit 230 in the present embodiment converts a typical authentication waveform and a typical registered waveform into a feature vector. The preliminary similarity calculation unit 230 changes the dimensions of the representative authentication waveform and the representative registered waveform to convert the representative authentication waveform and the representative registered waveform into a feature vector. For example, the preliminary similarity calculation unit 230 uses an AC / DCT, FTFT, MFCC, Wavelet, LPC / LPCC, or LSP to convert a representative authentication waveform and a representative registered waveform into a frequency domain. The preliminary similarity calculation unit 230 selects a characteristic component of a typical authentication waveform and a typical registered waveform converted into a frequency domain based on PCA, LDA, or ERE, and selects a representative authentication waveform and a typical authentication waveform. The dimension of the registered waveform is reduced, and a feature vector is generated from a typical authentication waveform with a reduced dimension and a typical registered waveform.

The preliminary similarity calculation unit 230 in the present embodiment extracts characteristic parameters of a typical authentication waveform and a typical registered waveform. For example, the preliminary similarity calculation unit 230 has a PR interval, a PR segment, a QRS complex, an ST segment, an ST interval, a QT interval, an RR interval, an amplitude, or a noise power of a typical authentication waveform and a typical registered waveform. At least one of is extracted as a feature parameter.

The preliminary similarity calculation unit 230 calculates the similarity between the characteristics of a typical authentication waveform and the characteristics of a typical registered waveform. For example, the preliminary similarity calculation unit 230 is representative of a feature vector in which a typical authentication waveform is converted using relevance, cosine similarity, Euclidean distance, L1 nom, P-nom, or root mean square error. The similarity between the registered waveform and the converted feature vector, or the similarity between the feature parameter of the typical authentication waveform and the feature parameter of the typical registered waveform is extracted. In addition, the preliminary similarity calculation unit 230 can extract the similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform by using all the methods capable of comparing the characteristics. ..

The preliminary similarity calculation unit 230 in the present embodiment shows the similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform in the form of a histogram, a feature vector, or a value.

The authentication unit 240 uses the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, and the similarity between the representative authentication waveform and the representative registered waveform. Determine if the user matches a user in the registered dataset.

In one embodiment, in order to perform authentication, the authentication unit 240 is registered with each of a plurality of extracted authentication waveforms using a normalization function such as min-max, Tanh-estimator, or z-score. Normalize the similarity between each of the waveforms and the similarity between a typical authentication waveform and a typical registered waveform.

Further, the authentication unit 240 determines the degree of similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, and the degree of similarity between the representative authentication waveform and the representative registered waveform. Authenticate the user by setting the weights differently. For example, the authentication unit 240 presets the weights, and sets the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, and representative authentication waveforms and representative registered waveforms. Set the weights by comparing the similarity between them. As an example, the authentication unit 240 averages the degree of similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, and between the representative authentication waveform and the representative registered waveform. A relatively high weight is given to a similarity having a high value among the similarities, and a relatively low weight is given to a similarity having a low value.

The authentication unit 240 in the present embodiment has the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms in the classifier, and the representative authentication waveform and the representative registered waveform. It is determined whether or not the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other by applying the similarity between them. As an example, the classifier is SVM or NN. When it is determined that the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other, the authentication unit 240 authenticates that the user of the authentication data set matches the user of the registered data set, and a plurality of extracted authentication waveforms are extracted. When it is determined that the authentication waveform and the plurality of registered waveforms do not correspond to each other, the authentication unit 240 authenticates that the user of the authentication data set does not match the user of the registered data set. Not only the similarity between each of the multiple extracted authentication waveforms and each of the multiple registered waveforms for user authentication, but also the similarity between the representative authentication waveform and the representative registered waveform. By applying both to the classifier, the false recognition rate and false rejection rate can be reduced.

When the authentication unit 240 determines that the plurality of extracted authentication waveforms and the plurality of registered waveforms do not correspond to each other, the authentication device 200 includes the data set generation unit 210, the similarity calculation unit 220, and the preliminary similarity calculation unit. The operations of 230 and the authentication unit 240 are repeated. In this case, when the number of repetitions of the operations of the data set generation unit 210, the similarity calculation unit 220, the preliminary similarity calculation unit 230, and the authentication unit 240 exceeds the predetermined critical number of repetitions, the authentication unit 240 Authenticate if the user does not match the user in the registration dataset.

3 and 4 are block diagrams showing an authentication device according to still another embodiment.

Referring to FIG. 3, the authentication device 300 includes a first feature extraction unit 310, a second feature extraction unit 320, a preliminary feature extraction unit 330, a matching unit 340, and an authentication unit 350. The first feature extraction unit 310, the second feature extraction unit 320, the preliminary feature extraction unit 330, the matching unit 340, and the authentication unit 350 include one or more processors or memories.

The authentication device 300 in this embodiment operates according to the operation mode. The operation mode includes a registration mode and an authentication mode. In the registration mode, the authentication device 300 performs an operation of registering the biometric signal of the registered user indicating the user to be registered, and in the authentication mode, the authentication device 300 uses the biometric signal of the authenticated user indicating the user to be authenticated. Performs an operation to authenticate the user.

Hereinafter, the authentication device 300 that operates in the authentication mode will be described.

The first feature extraction unit 310 extracts the features of a plurality of registered waveforms extracted from the registered user's electrocardiogram signal. The first feature extraction unit 310 extracts a plurality of registered waveforms from the registered user's electrocardiogram signals acquired in the registration mode. The first feature extraction unit 310 divides the electrocardiogram signal based on the heart rate and extracts P wave, QRS wave, T wave, and U wave from one registered waveform. The first feature extraction unit 310 includes a plurality of extracted registered waveforms in the registered data set.

In addition, the first feature extraction unit 310 extracts the features of each of the plurality of registered waveforms. As an example, the first feature extraction unit 310 extracts each feature vector or feature parameter of a plurality of registered waveforms.

The second feature extraction unit 320 extracts features of a plurality of authentication waveforms extracted from the electrocardiogram signal of the authenticated user. The second feature extraction unit 320 extracts a plurality of authentication waveforms from the electrocardiogram signal of the authenticated user acquired in the authentication mode. The second feature extraction unit 320 divides the electrocardiogram signal based on the heart rate and extracts the P wave, the QRS wave, the T wave, and the U wave from one registered waveform. The second feature extraction unit 320 includes a plurality of extracted authentication waveforms in the authentication data set.

In addition, the second feature extraction unit 320 extracts each feature (for example, a feature vector or a feature parameter) of the plurality of extracted authentication waveforms.

In the example shown in FIG. 3, the first feature extraction unit 310 and the second feature extraction unit 320 are expressed as being separated, but the first feature extraction unit 310 and the second feature extraction unit 320 are separated units. Or one unit.

The matching unit 340 matches each feature of the plurality of registered waveforms with each feature of the plurality of extracted authentication waveforms on a one-to-one basis, and performs one-to-one matching with each feature of the plurality of registered waveforms and a plurality of extracts. Calculate the similarity between each feature of the authentication waveform. When the plurality of extracted authentication waveforms is M and the plurality of registered waveforms included in the registered data set are N, the matching unit 340 performs M × N matching and M × N. Calculate the similarity.

As an example, matching unit 340 uses relevance, cosine similarity, Euclidean distance, L1 nom, P-nom, or root mean square error to convert a plurality of extracted authentication waveforms into a plurality of feature vectors. Similarity between each and each of the multiple registered waveforms converted, or each feature parameter of each of the multiple extracted authentication waveforms and each feature parameter of the plurality of registered waveforms. Extract the similarity between and.

The preliminary feature extraction unit 330 includes a representative registered waveform among a plurality of registered waveforms included in the registered data set and a representative authentication waveform among a plurality of extracted authentication waveforms included in the authentication data set. The features are extracted to extract the similarity between the features of the representative registered waveform and the features of the representative authentication waveform. In one embodiment, the average value of a plurality of registered waveforms included in the registered data set is calculated to extract a representative registered waveform, and the average value of the plurality of extracted authentication waveforms included in the authentication data set is calculated. Calculate and extract a typical authentication waveform. The preliminary feature extraction unit 330 extracts the features of a typical registered waveform (for example, a feature vector or a feature parameter) and the features of a typical authentication waveform.

In addition, the preliminary feature extraction unit 330 calculates the degree of similarity between the features of a typical registered waveform and the features of a typical authentication waveform. Since there is one typical registered waveform in the registered data set and one typical authentication waveform in the authentication data set, the preliminary feature extraction unit 330 calculates one similarity. As an example, the matching section uses the relevance, cosine similarity, Euclidean distance, L1 nom, P-nom, or root mean square error to form a typical registered waveform feature vector and a typical certified waveform feature vector. The similarity between, or the similarity between the characteristic parameter of a typical registered waveform and the characteristic parameter of a typical authentication waveform is extracted.

The authentication unit 350 uses the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. Determine if the user matches a user in the registered dataset.

The authentication unit 350 in the present embodiment has a similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms, and a similarity between the representative authentication waveform and the representative registered waveform. Normalize the degree and weight the similarity between each of the multiple extracted authentication waveforms and each of the multiple registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. Set differently.

The authentication unit 350 describes the similarity between each of the plurality of extracted authentication waveforms and each of the plurality of registered waveforms in the classifier (for example, SVM, NN), and the representative authentication waveform and the representative registration. The similarity between the waveforms is applied to determine whether or not the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other.

When it is determined that the plurality of extracted authentication waveforms and the plurality of registered waveforms correspond to each other, the authentication unit 350 authenticates that the user of the authentication data set matches the user of the registered data set, and multiple extracted authentications. When it is determined that the waveform and the plurality of registered waveforms do not correspond to each other, the authentication unit 350 authenticates that the user of the authentication data set does not match the user of the registered data set.

In another embodiment, the authentication unit 350 applies a similarity between each of the plurality of registered waveforms and each of the plurality of extracted authentication waveforms to the classifier, and then the representative registered waveform and the representative. After applying the similarity between the two registered waveforms and the similarity between the representative registered waveform and the typical certified waveform to the classifier, each of the multiple registered waveforms and multiple extracted waveforms. Apply the similarity between each of the certified waveforms. For example, the authentication unit 350 applies the similarity between each of the plurality of registered waveforms and each of the plurality of extracted authentication waveforms in the classifier to correspond the plurality of registered waveforms with the plurality of authentication waveforms. If it is determined whether or not to do so and it is determined that multiple registered waveforms correspond to multiple certified waveforms, the similarity between the typical registered waveform and the typical certified waveform is applied to the classifier. It is determined whether or not a typical registered waveform and a typical authentication waveform correspond to each other. Here, a plurality of registered waveforms and a plurality of extracted authentication waveforms are obtained by applying the similarity between each of the plurality of registered waveforms and each of the plurality of extracted authentication waveforms in the classifier. If it is determined that it does not correspond, the authentication unit 350 authenticates that the user does not match the user in the registered data set.

As another example, does the authentication unit 350 apply the similarity between the representative registration waveform and the representative authentication waveform to the classifier so that the representative registration waveform and the representative authentication waveform correspond to each other? When it is determined whether or not the representative registered waveform and the representative authentication waveform correspond to each other, the authentication unit 350 determines each of the plurality of registered waveforms in the classifier and the plurality of extracted authentication waveforms. By applying the similarity between each of the above, it is determined whether or not the plurality of registered waveforms and the plurality of extracted authentication waveforms correspond to each other. When it is determined that the plurality of registered waveforms and the plurality of extracted authentication waveforms correspond to each other, the authentication unit 350 authenticates that the user matches the user of the registered data set.

For user authentication, not only the similarity between each of the multiple extracted authentication waveforms and each of the multiple registered waveforms, but also the similarity between the representative authentication waveform and the representative registered waveform. By using both for user authentication, the false recognition rate or false rejection rate can be reduced.

Referring to FIG. 4, the authentication device 400 includes a sensor 410, a memory 420, a processor 430, and a display 440. Each of the sensor 410, the memory 420, the processor 430, and the display 440 includes one or more hardware components.

The sensor 410 includes an electrocardiogram sensor, an electrocardiogram sensor, an electromyogram sensor, and / or an electrocardiogram sensor. Also, the sensor 410 may include different sensors capable of extracting signals sensed from the human body. When the sensor 410 includes an electrocardiogram sensor, the electrocardiogram sensor measures the user's electrocardiogram waveform using the first electrode, the second electrode, and the third electrode. The electrocardiogram sensor in this embodiment includes first to third electrodes, an amplifier, and a digital converter. The first electrode to the third electrode come into contact with the user's skin and measure the user's electrocardiogram signal. The amplifier amplifies the electrocardiogram signal measured at the first electrode to the third electrode. The digital converter converts the amplified electrocardiogram signal into a digital signal and extracts the electrocardiogram waveform.

The memory 420 stores in advance a registration data set including a plurality of registered waveforms extracted from the biomedical signals of the user registered in advance.

The processor 430 extracts a plurality of authentication waveforms for user authentication from the biometric signal to generate an authentication data set, and each of the plurality of extracted authentication waveforms is included in the pre-stored registration data set. The similarity between each of the plurality of extracted authentication waveforms and the plurality of registered waveforms is calculated by one-to-one matching with the registered waveforms of.

The processor 430 in the present embodiment extracts each feature of the plurality of extracted authentication waveforms and each feature of the plurality of registered waveforms, and each feature of the plurality of extracted authentication waveforms and a plurality of registered ones. Calculate the similarity between each feature of the waveform. Further, the processor 430 extracts the characteristics of the representative authentication waveform indicating the representative waveform of the plurality of extracted authentication waveforms and the representative registered waveform indicating the representative waveform of the plurality of registered waveforms, and performs the representative authentication. Calculate the similarity between waveform features and typical registered waveform features.

The processor 430 has a similarity between each feature of the plurality of extracted authentication waveforms and each feature of the plurality of registered waveforms, or a representative authentication waveform feature and a representative registered waveform feature. Authenticate the user with at least one of the similarities between them.

The processor 430 also displays information about user authentication on the display 440. For example, the processor 430 displays information on the user's biological signal, information on the extracted authentication waveform, information on the registered waveform, or information on the user's authentication result on the display 440.

The processor 430 includes a data set generation unit 110 shown in FIG. 1, a similarity calculation unit 120, an authentication unit 130, a data set generation unit 210 shown in FIG. 2, a similarity calculation unit 220, a preliminary similarity calculation unit 230, and an authentication unit. 240, since the contents explained by the first feature extraction unit 310, the second feature extraction unit 320, the preliminary feature extraction unit 330, the matching unit 340, and the authentication unit 350 shown in FIG. 3 are applied as they are, a more detailed explanation is omitted. To do.

5 (a) and 5 (b) are diagrams for explaining user authentication of the authentication device according to one embodiment.

Referring to FIG. 5, FIGS. 5A and 5B show ECG waveforms extracted from the ECG signal of the authenticating user. The horizontal axis of FIGS. 5A and 5B shows time, and the vertical axis shows the magnitude of the electrocardiogram waveform.

FIG. 5A shows an electrocardiogram waveform when the user's breathing is stable. The authentication device divides the electrocardiogram waveform of FIG. 5A based on the heart rate, extracts each of the electrocardiogram waveforms 511 to 519, and stores each of the electrocardiogram waveforms 511 to 519 in the authentication data set stored in the memory. To do. The authentication device matches each of the electrocardiogram waveforms 511 to 519 one-to-one with a plurality of electrocardiogram waveforms included in the registered data set stored in the memory in advance, and includes each of the electrocardiogram waveforms 511 to 519 and the registered data set. The similarity between a plurality of electrocardiogram waveforms is calculated, and the user is authenticated using the similarity.

FIG. 5B shows an electrocardiogram waveform when the user's breathing becomes unstable after it stabilizes. The user's breathing is stable in section 520 and the user's breathing is unstable in section 530.

The authentication device divides the electrocardiogram waveform of FIG. 5B based on the heart rate and extracts the electrocardiogram waveforms 541 to 549. The deviation of the electrocardiogram waveforms 541 to 546 is relatively small due to the stabilization of the user's breathing, but the deviation of the electrocardiogram waveforms 547 to 549 is relatively large due to the unstable breathing of the user. The authentication device stores each of the electrocardiogram waveforms 541 to 549 in the authentication data set. The authentication device authenticates the user using the electrocardiogram waveforms 541 to 549. Here, if the authentication device extracts a representative electrocardiogram waveform from the electrocardiogram waveforms 541 to 549 and then authenticates the user using only the representative electrocardiogram waveform, the deviation from the other electrocardiogram waveforms 541 to 546 is large. Typical electrocardiogram waveforms are distorted by the electrocardiogram waveforms 547 to 549, which increases the false recognition rate or false rejection rate. Therefore, in order to reduce the false recognition rate or false rejection rate, the authentication device matches each of the electrocardiogram waveforms 541 to 549 one-to-one with a plurality of electrocardiogram waveforms included in the pre-stored registered data set, and performs an electrocardiogram. The similarity between each of the waveforms 541 to 549 and the plurality of electrocardiogram waveforms included in the registered data set is calculated, and the user is authenticated using the similarity.

FIG. 6 is a diagram for explaining the calculation of the degree of similarity between the extracted authentication waveform and the registered waveform according to the embodiment.

Referring to FIG. 6, the electrocardiogram waveform 610 shows the registered waveform included in the registered data set, and the electrocardiogram waveform 650 shows the extracted authentication waveform included in the authentication data set.

The authentication device converts the registered waveform 610 into the frequency domain. As an example, the authentication device converts the waveform 610 registered using the FTFT into the frequency domain. The authentication device also reduces the dimension of the registered waveform 620 converted to the frequency domain to generate a feature vector. As an example, the authentication device selects the feature components 631, 632, 633, 634, 635 in the registered waveform 620 converted into the frequency domain using the LDA, and converts the registered waveform 620 into the frequency domain. The dimension is reduced and the feature vector 681 is acquired from the registered waveform 630 with the reduced dimension.

The authentication device also converts the extracted authentication waveform 650 into the frequency domain. As an example, the authentication device converts the authentication waveform 650 extracted using the FTFT into the frequency domain. Further, the authentication device selects the feature components 671, 672, 673, 674, and 675 from the extracted authentication waveform 660 converted into the frequency domain using the LDA, and converts the extracted authentication waveform 660 into the frequency domain. The feature vector 682 is acquired from the extracted authentication waveform 670 with the reduced dimension.

The authentication device calculates the similarity between the feature vector 681 acquired from the registered waveform 610 and the feature vector 682 acquired from the extracted authentication waveform 650. For example, the authentication device uses the cosine similarity to calculate the similarity 683 between the feature vector 681 and the feature vector 682.

FIG. 7 is a diagram for explaining matching between the extracted authentication waveform and the registered waveform according to the embodiment.

Referring to FIG. 7, the authentication apparatus includes the three extracted authentication waveforms 711, 712, 713 contained in the authentication data set 710 and the four registered waveforms 721, 722, 723 contained in the registration data set 720. , 724 are matched one-to-one with each of the above, and matching is performed 12 times. In FIG. 7, the extracted authentication waveforms 711, 712, 713 and the registered waveforms 721, 722, 723, 724 indicate feature vectors or feature parameters. By matching, the authentication device calculates the similarity between each of the extracted authentication waveforms 711, 712, 713 and each of the registered waveforms 721, 722, 723, 724. For example, in the case of the extracted authentication waveform 711, the authentication device has a similarity between the extracted authentication waveform 711 and the registered waveform 721, and between the extracted authentication waveform 711 and the registered waveform 722. The similarity, the similarity between the extracted authentication waveform 711 and the registered waveform 723, and the similarity between the extracted authentication waveform 711 and the registered waveform 724 are calculated.

The authentication device in this embodiment is registered with each of the extracted authentication waveforms 711, 712, 713 using relevance, cosine similarity, Euclidean distance, L1 nom, P-nom, or root mean square error. The similarity between the waveforms 721, 722, 723, and 724 is calculated.

FIG. 8 is a diagram for explaining an example of an authentication device according to an embodiment.

Referring to FIG. 8, the mobile terminal 810 includes an anode electrode 821, a reference electrode 822, and a cathode electrode 830 for sensing an electrocardiogram signal. The anode electrode 821 and the reference electrode 822 in the present embodiment are located on the side surface of the mobile terminal 810, and the cathode electrode 830 is located on the lower stage of the mobile terminal 810.

When the user touches a plurality of electrodes (821, 822, 830) with a finger, the mobile terminal 810 senses an electrocardiogram signal. The mobile terminal 810 amplifies the electrocardiogram signal using an amplifier and converts it into a digital signal using a digital filter. The mobile terminal 810 extracts an electrocardiogram waveform from the converted digital signal.

When the electrocardiogram waveform is registered, the mobile terminal 810 acquires the user's electrocardiogram signal using a plurality of electrodes (821, 822, 830) during the registration, and obtains the acquired electrocardiogram signal based on the heart rate. Divide and extract multiple waveforms for registration. The mobile terminal 810 includes the extracted waveform in the registered data set. The mobile terminal 810 stores the registered data set in the memory.

When the user authenticates, the mobile terminal 810 acquires the user's electrocardiogram signal using a plurality of electrodes (821, 822, 830), divides the acquired electrocardiogram signal based on the heart rate, and performs a plurality of authentications. Extract the waveform. The mobile terminal 810 includes a plurality of extracted authentication waveforms in the authentication data set. The mobile terminal 810 matches each of the plurality of extracted authentication waveforms included in the authentication data set with the plurality of registered waveforms included in the registered data set on a one-to-one basis with each of the plurality of extracted authentication waveforms. Calculate the similarity between multiple registered waveforms. Further, the mobile terminal 810 extracts a representative authentication waveform from a plurality of extracted authentication waveforms, extracts a representative registered waveform from the plurality of registered waveforms, and extracts a representative authentication waveform and a representative registration waveform. Calculate the similarity with the waveform. The mobile terminal 810 is authenticated by using the similarity between each of the plurality of extracted authentication waveforms and the plurality of registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. Authenticates whether or not the user who performs the above matches the user registered in advance.

FIG. 9 is a diagram for explaining an example of an authentication device according to another embodiment.

Referring to FIG. 9, the wearable terminal 910 includes an anode electrode 921, a reference electrode 922, and a cathode electrode 931 for sensing an electrocardiogram signal. In one embodiment, the anode electrode 921 and the reference electrode 922 are located on the rear surface of the wearable terminal 910, and the cathode electrode 931 is located on the front surface of the wearable terminal 99.

Similar to the mobile terminal 810 shown in FIG. 8, the wearable terminal 910 acquires a user's electrocardiogram signal using a plurality of electrodes (921, 922, 931), extracts an electrocardiogram waveform from the electrocardiogram signal, and extracts the electrocardiogram waveform. Filter the electrocardiogram waveform. Further, the wearable terminal 910 performs the same operation as the mobile terminal 810 shown in FIG. 8, and either registers the electrocardiogram waveform of the user in advance or authenticates whether or not the user to be authenticated matches the registered user in advance. To do.

FIG. 10 is a diagram for explaining an example of an authentication device according to still another embodiment. The authentication device of FIG. 10 is a mobile terminal.

Referring to FIG. 10, the mobile terminal 1010 includes an anode electrode 1021 for sensing an electrocardiogram signal, a reference electrode 1022, and a cathode electrode 1030. The anode electrode 1021 and the reference electrode 1022 in the present embodiment are located on the side surface of the mobile terminal 1010, and the cathode electrode 1030 is located on the lower stage of the mobile terminal 1010.

The mobile terminal 1010 acquires a user's electrocardiogram signal using a plurality of electrodes (1021, 1022, 1030), divides the acquired electrocardiogram signal based on the heart rate, and extracts a plurality of authentication waveforms. The mobile terminal 1010 includes a plurality of extracted authentication waveforms in the authentication data set.

The mobile terminal 1010 in the present embodiment receives information about the registration data set from the server 1040 by the communication interface. Here, the communication interface, WLAN (Wireless LAN), WiFi (Wireless Fidelity) Direct, DLNA (Digital Living Network Alliance), Wibro (Wireless broadband), Wimax (World Interoperability for Microwave Access), HSDPA (High Speed Downlink Packet Access ) And other wireless internet interfaces, Bluetooth (Bluetooth®), RFID (Radio WiFi Identification), Infrared Data Association (IrDA), UWB (Ultra WideBand), ZigBe Includes short-range communication interfaces such as Communication). Not only that, the communication interface may include any interface (eg, a wired interface) capable of communicating with the server 1040.

The mobile terminal 1010 matches each of the plurality of extracted authentication waveforms included in the authentication data set with the plurality of registered waveforms included in the registered data set on a one-to-one basis with each of the plurality of extracted authentication waveforms. Calculate the similarity between multiple registered waveforms. Further, the mobile terminal 1010 extracts a representative authentication waveform from a plurality of extracted authentication waveforms, extracts a representative registered waveform from the plurality of registered waveforms, and performs a representative authentication waveform and a representative registration. Calculate the similarity with the waveform. The mobile terminal 1010 is authenticated by using the similarity between each of the plurality of extracted authentication waveforms and the plurality of registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. Authenticates whether or not the user who performs the above matches the user registered in advance.

The mobile terminal 1010 transmits information on the user's electrocardiogram signal, information on the extracted authentication waveform, or information on the user's authentication result to the server 1040 by the communication interface. The server 1040 in the present embodiment allows the user to access the server 1040 by using the information regarding the authentication result received from the mobile terminal 1010.

FIG. 11 is an operation flowchart showing an authentication method according to one embodiment.

Referring to FIG. 11, the authentication device extracts a plurality of authentication waveforms for user authentication from the user's biological signal to generate an authentication data set (step S1110).

Next, the authentication device matches each of the plurality of extracted authentication waveforms with the plurality of registered waveforms included in the pre-stored registration data set on a one-to-one basis, and obtains the plurality of extracted authentication waveforms. The similarity between each and the plurality of registered waveforms is calculated (step S1120).

Next, the authentication device extracts a representative authentication waveform showing a representative waveform of a plurality of extracted authentication waveforms and a representative registered waveform showing a representative waveform of a plurality of registered waveforms, and a representative authentication waveform. The degree of similarity between and the representative registered waveform is calculated (step S1130).

Finally, the authentication device uses the similarity between each of the plurality of extracted authentication waveforms and the plurality of registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. Authenticate the user (step S1140).

Since the contents described with reference to FIGS. 1 to 10 are applied as they are to the authentication method according to the embodiment shown in FIG. 11, a more detailed description will be omitted.

The above-mentioned device is embodied by a hardware component, a software component, or a combination of a hardware component and a software component. For example, the apparatus and components described in this embodiment include, for example, a processor, a controller, an ALU (arithmetic logic unit), a digital signal processor, a microprocessor, an FPA (field program array), and a PLU (programmable). It is embodied using one or more general purpose computers or special purpose computers, such as units, microprocessors, or all different devices that execute and respond to instructions. The processing device executes an operating system (OS) and one or more software applications running on the operating system. The processing device also accesses, stores, manipulates, processes, and generates data in response to software execution. For convenience of understanding, it may be explained that one processing device is used, but those who have ordinary knowledge in the technical field may explain that the processing device has multiple processing elements or multiple types. It can be seen that it contains the processing elements of. For example, the processing device includes a plurality of processors or one processor and one controller. In addition, other processing configurations such as a parallel processor are also possible.

The software includes computer programs, codes, instructions, or a combination of one or more of them, and configures the processing device to operate as desired, and the processing device is independent or collectible. To order. The software and / or data is interpreted by the processing device or transmitted to the processing device in order to provide instructions or data to all types of machines, components, physical devices, virtual devices, computer storage media or devices, or transmissions. It is embodied permanently or temporarily by the signal wave to be generated. The software is distributed on a networked computer system and is stored or executed in a distributed manner. The software and data are stored on one or more computer-readable recording media.

The method according to this embodiment is embodied in the form of program instructions executed via various computer means and recorded on a computer-readable recording medium. Computer-readable recording media include program instructions, data files, data structures, etc., alone or in combination. The program instructions recorded on the recording medium may be those specially designed and configured for this embodiment, or those known and used by those skilled in the art of computer software. Examples of computer-readable recording media include hard disks, floppy (registered trademark) disks, magnetic media such as magnetic tapes, optical recording media such as CD-ROMs and DVDs, and floptic discs and optical discs. Optical porcelain media such as, and hardware devices specially configured to store and execute program instructions such as ROM, RAM, flash memory, etc. are included. Examples of program instructions include not only machine language code as created by a compiler, but also high-level language code executed by a computer using an interpreter or the like. Hardware devices may be configured to operate as one or more software modules to perform the operations of the present invention, and vice versa.

Although the embodiments of the present invention have been described in detail with reference to the drawings, the present invention is not limited to the above-described embodiments and is variously modified within a range that does not deviate from the technical scope of the present invention. It is possible to carry out.

100, 200, 300, 400 Authentication device 110, 210 Data set generation unit 120, 220 Similarity calculation unit 130, 240, 350 Authentication unit 230 Preliminary similarity calculation unit 310 First feature extraction unit 320 Second feature extraction unit 330 Spare Feature Extraction unit 340 Matching unit 410 Sensor 420 Memory 430 Processor 440 Display 511 to 519, 541 to 549 Electrocardiogram waveform 520, 530 Section 610, 620, 630, 721 to 724 Registered waveforms 650, 660, 711 to 713 Extracted Authentication waveform 613-635, 671-675 Feature component 681, 682 Feature vector 710 Authentication data set 720 Registration data set 810, 1010 Mobile terminal 821, 921, 1021 Anopole electrode 822, 922, 1022 Reference electrode 830, 931, 1030 Cone electrode 910 Wearable terminal 1040 server

Claims (23)

A data set generator that extracts multiple authentication waveforms from the user's biological signal and generates an authentication data set,
Each of the extracted authentication waveforms is matched with each of the plurality of registered waveforms included in the registered data set stored in the memory, and each of the extracted authentication waveforms and each of the registered waveforms is used. A similarity calculation unit that calculates the similarity between
A representative authentication waveform showing a representative waveform of the extracted authentication waveform and a representative registered waveform showing a representative waveform of the registered waveform are extracted to form the representative authentication waveform and the representative registered waveform. A preliminary similarity calculation unit that calculates the similarity between
Authentication that authenticates a user using the similarity between each of the extracted authentication waveforms and each of the registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. With a department,
When the extracted authentication waveforms are M and the plurality of registered waveforms are N, the similarity calculation unit performs M × N matching and M extracted authentication waveforms. authentication apparatus characterized that you calculating the similarity between each of the respective N number of registered waveform. The memory is characterized in that it stores a registration data set including a plurality of registered waveforms extracted from a user's biological signal during registration before the data set generation unit generates the authentication data set. The authentication device according to claim 1. The similarity calculation unit is
Each feature of the extracted authentication waveform and each feature of the registered waveform are extracted.
The authentication device according to claim 1, wherein the degree of similarity between each feature of the extracted authentication waveform and each feature of the registered waveform is calculated. The similarity calculation unit is
Each of the extracted authentication waveform and the registered waveform is converted into a feature vector, and then
The authentication apparatus according to claim 3, wherein the degree of similarity between each of the feature vectors to which the extracted authentication waveform is converted and each of the feature vectors to which the registered waveform is converted is calculated. .. The similarity calculation unit reduces the respective dimensions of the extracted authentication waveform and the registered waveform to convert each of the extracted authentication waveform and the registered waveform into the feature vector. The authentication device according to claim 4, wherein the authentication device is characterized. The similarity calculation unit is
Each feature parameter of the extracted authentication waveform and each feature parameter of the registered waveform are extracted.
The authentication device according to claim 3, wherein the degree of similarity between each feature parameter of the extracted authentication waveform and each feature parameter of the registered waveform is calculated. The characteristic parameter is at least one of the PR interval, PR segment, QRS complex, ST segment, ST interval, QT interval, RR interval, amplitude, or noise power of each of the extracted authentication waveform or the registered waveform. The authentication device according to claim 6, wherein the authentication device includes one. The preliminary similarity calculation unit is
The features of the typical authentication waveform and the typical registered waveform are extracted.
The authentication device according to claim 1, wherein the degree of similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform is calculated. The preliminary similarity calculation unit is
Each of the representative authentication waveform and the representative registered waveform is converted into a feature vector, and then
The authentication device according to claim 8, wherein the degree of similarity between the feature vector to which the representative authentication waveform is converted and the feature vector to which the representative registered waveform is converted is calculated. The preliminary similarity calculation unit is
The characteristic parameters of the typical authentication waveform and the characteristic parameters of the typical registered waveform are extracted.
The authentication device according to claim 8, wherein the degree of similarity between the characteristic parameter of the representative authentication waveform and the characteristic parameter of the representative registered waveform is calculated. The authentication unit has a degree of similarity between each of the extracted authentication waveforms and each of the registered waveforms, and similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. The authentication device according to claim 8, wherein the degree is used to determine whether or not the user matches the user of the registered data set. The authentication unit has a degree of similarity between each of the extracted authentication waveforms and each of the registered waveforms, and similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. The authentication device according to claim 11, wherein the degree is normalized. The authentication unit has a degree of similarity between each of the extracted authentication waveforms and each of the registered waveforms, and similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform. The authentication device according to claim 11, wherein the weights of the degrees are set differently to determine whether or not the user matches the user of the registered data set. The authentication unit has a similarity between each of the extracted authentication waveforms and each of the registered waveforms in the classifier, and features of the representative authentication waveform and features of the representative registered waveform. The authentication device according to claim 11, wherein it is determined whether or not the extracted authentication waveform and the registered waveform correspond to each other by applying the similarity between the two. The authentication device according to claim 14, wherein the classifier includes at least one of SVM (Support Vector Machine) and NN (Nearest Neighbor). When the authentication unit determines that the user does not match the user of the registered data set, the operations of the data set generation unit, the similarity calculation unit, the preliminary similarity calculation unit, and the authentication unit are repeated. The authentication device according to claim 11. When the number of repetitions of the operations of the data set generation unit, the similarity calculation unit, the preliminary similarity calculation unit, and the authentication unit exceeds a predetermined critical number of repetitions, the authentication unit causes the user to use the authentication unit. The authentication device according to claim 16, wherein it is determined that the user does not match the user of the registered data set. A sensor that senses the user's biological signal,
A plurality of authentication waveforms are extracted from the biological signal to generate an authentication data set, and the authentication data set is generated.
Each of the plurality of extracted authentication waveforms included in the generated authentication data set is matched with each of the plurality of registered waveforms stored in the memory, and each of the extracted authentication waveforms and the registered waveforms are registered. Calculate the similarity between each of the waveforms and
A representative authentication waveform showing a representative waveform of the extracted authentication waveform and a representative registered waveform showing a representative waveform of the registered waveform are extracted to form the representative authentication waveform and the representative registered waveform. Calculate the similarity between
A processor that authenticates a user using the similarity between each of the extracted authentication waveforms and each of the registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. And with
When calculating the similarity, when each of the extracted authentication waveforms is M and the plurality of registered waveforms are N, M × N matching is performed and M are extracted. authentication apparatus characterized that you calculating the similarity between each of the respective N number of registered waveform authentication waveform. The authentication device according to claim 1 or 18, wherein the biological signal includes an electrocardiogram signal. The processor
Each feature of the extracted authentication waveform and each feature of the registered waveform are extracted.
The authentication apparatus according to claim 18, wherein the degree of similarity between each feature of the extracted authentication waveform and each feature of the registered waveform is calculated. The processor
The features of the typical authentication waveform and the typical registered waveform are extracted.
The authentication apparatus according to claim 18, wherein the degree of similarity between the characteristics of the representative authentication waveform and the characteristics of the representative registered waveform is calculated. It is an authentication method of the user authentication device.
Steps to extract multiple authentication waveforms from the user's biological signal to generate an authentication dataset,
Each of the plurality of extracted authentication waveforms included in the generated authentication data set is matched with each of the plurality of registered waveforms included in the registered data set stored in the memory of the extracted authentication waveform. A step of calculating the similarity between each and each of the registered waveforms,
A representative authentication waveform showing a representative waveform of the extracted authentication waveform and a representative registered waveform showing a representative waveform of the registered waveform are extracted to form the representative authentication waveform and the representative registered waveform. And the steps to calculate the similarity between
A step of authenticating a user using the similarity between each of the extracted authentication waveforms and each of the registered waveforms and the similarity between the representative authentication waveform and the representative registered waveform. And have
The step of calculating the degree of similarity, the authentication waveform said extracted is M and the plurality of registered waveform when N pieces performs M × N times matching is the M extract authentication wherein that you calculating the similarity between each of the respective N number of registered waveform authentication waveform. A computer-readable recording medium on which a program for causing an authentication device to execute the authentication method according to claim 22 is recorded.

Images